外文翻譯--- - 傳感器和傳感器激勵和測量技術(shù)

1、<p><b>　　附錄A</b></p><p>　　Transducer and sensor excitation and measurement techniques</p><p>　　Many of today's industrial and instrumentation applications involving sensor.

2、The function of the sensor system is to monitor changes, and then this data back to the main controller. For a simple voltage or current measurement sensors may be resistance in nature. However, some sensor system may be

3、 inductive or capacitive in nature, that is to say, the frequency range of the sensor resistance change is nonlinear. Impedance sensors such typical example is the proximity sensor - a campaign for the </p><p&

4、gt;　　Inspired by the frequency of sensor signals based on sensor values of L or C to show the corresponding instantaneous magnitude, frequency or phase changes. For example, the ultrasound will show a flow of phase offse

5、t, while the proximity sensor will cause the rate to change. Tracking changes in impedance that is the most commonly used to monitor the resonant frequency circuit. Capacitance value of the resonant frequency is equal to

6、 the frequency where the inductance value point. This is also the</p><p>　　Calculated resonant frequency: calculation circuit measuring the resonant frequency of the need for the relationship between frequen

7、cy and impedance, in particular, need a certain frequency range with the ability to scan the waveform generator. A simple, low-cost method is based on the AD5930 waveform generator. AD5930 with a group of pre-set in the

8、frequency range of the ability to provide a linear scan. Once the conditions for setting, on the need for further control, in addition to a frequenc</p><p>　　AD5930 has many advantages: the output frequency

9、resolution of 28 bit, so you can be less than the control accuracy of 0.1 Hz output frequency. The output frequency range of 0 ~ 10 MHz, thus the selection of sensors with a lot of flexibility. For example, some sensors

10、a very narrow frequency range, but the requirements in this frequency range with high resolution. Some sensors may require a wide frequency range, but lower resolution requirements. This approach is easy to calculate the

11、 resonant f</p><p>　　System block diagram: typical block diagram of such a system as shown in Figure 3. Through the BF-535 DSP processor AD5930 digital waveform generator set. AD5930 needs arising from the s

12、inusoidal output voltage waveform for low-pass filtering and amplification in order to eliminate the master clock (MCLK), mirroring the frequency and high frequency noise generated by feed through. After filtering the se

13、nsor signal can be used as a source of excitation frequency. According to the impedance of the</p><p>　　Complete integrated sensor solutions: separation described above is a common solution for impedance mea

14、surement of the sensor solution. The program may require many discrete components, so the sensor is a cost-analysis solution. These separate components will increase their own sources of error. The design of active compo

15、nents will increase the number of phase error, which is the need for correction. In addition, the DSP also need to deal with some complex mathematical calculations, this may req</p><p>　　Address the above-me

16、ntioned analysis of the issue of low-frequency sensor solutions AD5933 / 4 device, it will by the main processing module are integrated into one chip. The core of the chip, including three main modules: the frequency of

17、scanning for the direct digital synthesizer (DDS) waveform generator; used to measure the sensor's response to the 12 bit, 1 MSPS ADC; and, finally, to the ADC data for 1024 point Discrete Fourier Transform (DFT) cal

18、culations of the DSP engine. </p><p>　　The results of DFT calculations to provide a real part (R) and an imaginary part (I) data, which can easily calculate the impedance. Using the following formula is easy

19、 to calculate the impedance amplitude and phase: </p><p>　　In order to determine the actual value of the real impedance Z (ω), is typically required to perform frequency scanning. Can calculate the impedance

20、 of each frequency point, which can draw a relationship between frequency and amplitude curves. So it is easy to measure 100 Ω ~ 20 MΩ resistance within the scope. The system allows users to set up a 2 V peak-to-peak (PK

21、-PK) of the sinusoidal signal as an external frequency source excitation load. Output range can be set to 1V, 500 mV and 200 mV. Fre</p><p>　　In order to achieve the frequency of scanning, the user must firs

22、t set up the required frequency of scanning conditions: the need for a start frequency, frequency interval and sweep points. Then the need for a start command to start scanning. Frequency points in each scan, ADC complet

23、ed the first 1024 samples, and then calculating the DFT in order to provide the waveform of the real and imaginary parts of the data. The real and imaginary parts of the data through the I2C interface in the form of</

24、p><p>　　Without calibration of the system as a result can only use the typical value of sensitivity and offset the output voltage is converted to pressure, the pressure measured will have a margin of error as s

25、hown in Figure .</p><p>　　This initial error without calibration by the following components:</p><p>　　1.offset error: As the pressure in the entire range of vertical shift to maintain a constan

26、t, so the proliferation and laser conditioning converter changes the amendment would have offset error.</p><p>　　2.The sensitivity of error, resulting in errors in direct proportion to the size and pressure:

27、 If the device is higher than the typical value of the sensitivity, the sensitivity of the error will be incremental pressure function (see Figure 1). If the sensitivity is lower than the typical value, then the sensitiv

28、ity of the error will be decreasing function of pressure. The cause of the error diffusion process is to change.</p><p>　　3.Linearity Error: This is an initial error factor less affected, the error is the ca

29、use of the physical non-linear silicon, but with the sensor amplifier, should also include non-linear amplifier. Linear error curve can be concave curve, it could be a convex curve. </p><p>　　4.Lag Error: In

30、 most cases, the lag error can be ignored completely, because silicon has a higher degree of mechanical stiffness. Changes in general just a lot of pressure to consider the case of hysteresis error.</p><p>　

31、　Calibration can eliminate or greatly reduce these errors, and compensation technique is usually required to identify the parameters of the actual transfer function, rather than simply the use of typical values. Potentio

32、meter, adjustable resistance, and other hardware can be used in the compensation process, while the software is able to achieve more flexibility in the work of this error compensation. </p><p>　　Calibration

33、method that can eliminate the transfer function against the Agency to compensate the offset drift error, such as the auto-zero calibration method. Offset zero calibration is usually carried out under pressure, especially

34、 in the differential sensor, because under the conditions of the nominal differential pressure is usually 0. For pure sensor offset calibration will be difficult, because it either needs to read a pressure system to meas

35、ure the atmospheric pressure in the environment </p><p>　　Calibration is very important selection pressure, which determines the accuracy to obtain the best pressure range. In fact, after calibration offset

36、actual standard fixed-point error in the Department and has been to maintain a smaller minimum value. Therefore, the reference points must be in accordance with the scope of the target selection pressure, and pressure ra

37、nge can not be consistent with the scope of work. </p><p>　　In order to convert the pressure of the output voltage value, usually as a result of the actual sensitivity is unknown, and therefore the mathemati

38、cal model used for a typical single-point calibration sensitivity.</p><p>　　Said that the red curve calibration offset (PCAL = 0) after the error curve, the error can be found that calibration curve relative

39、 to the black before the error had a vertical offset curve.</p><p>　　This calibration method and calibration method that is more stringent requirements to achieve a higher cost. However, compared with the ca

40、libration point, the method can significantly improve the accuracy of the system, because the method is not only an offset calibration, the calibration of the sensor sensitivity. Therefore, the calculation error can be u

41、sed in the actual value of sensitivity, and the atypical values.       That improve the accuracy of the green curve. Here, calibration is tri</p><p>　　Some applications require

42、 the pressure in the whole range of high accuracy. In these applications, can be used multi-point calibration method to get the best results. In multi-point calibration method, not only considered the error of offset and

43、 sensitivity, but also takes into account most of the linear error curve shown in purple. The mathematical model used here, with each calibration interval (between the two reference points) exactly the same as a two-tier

44、 calibration. </p><p>　　As mentioned earlier, the linear form of a consistent error and the error curve in line with the quadratic equation of the curve, with a predictable size and shape. Did not use the am

45、plifier for the sensor, especially because of the nonlinear sensor is based on the nature of mechanical reasons (this is caused by the pressure of silicon thin-film). Linear description of the error characteristics of a

46、 typical example can be calculated the average linear error to determine the polynomial function </p><p>　　Examples of compensation MPX2300 Motorola, MPX2300 is a blood pressure measurement is mainly used in

47、 the temperature compensation sensor. Polynomial model can be an average of 10 sensors to be linear error compensation of the error after the initial maximum linearity error of about ten to one-twentieth, as shown in dot

48、ted line in Figure 3. The error compensation method can be only two points calibration for high-performance low-cost sensors to improve the device (full scale error of less than 0.</p><p><b>　　附錄B</

49、b></p><p>　　傳感器和傳感器激勵和測量技術(shù)</p><p>　　當今的許多工業(yè)和儀器儀表應用都涉及傳感器測量。傳感器的功能就是監(jiān)視系統(tǒng)中的變化，然后將此數(shù)據(jù)反饋給主控制器。用于簡單的電壓或電流測量的傳感器可能是電阻性的。但是，有些傳感器系統(tǒng)可能是電感性或電容性的，就是說在傳感器頻率范圍內(nèi)阻抗變化是非線性的。 這類復阻抗傳感器的典型例子就是接近傳感器——用于檢測一個運動物體

50、的相對距離；另外，容性傳感器或感性傳感器——在醫(yī)用設(shè)備中用于測量血流或者分析血壓或血質(zhì)。 為了用這些“復阻抗傳感器”實現(xiàn)測量，必須提供一種交流（AC）激勵頻率源在傳感器的頻率范圍內(nèi)進行掃描。本文試圖說明如何采用單芯片數(shù)字波形發(fā)生器輕松實現(xiàn)這種高達10 MHz的頻率掃描。還介紹了一種帶集成激勵、響應和數(shù)字信號處理器（DSP）功能完整的單芯片傳感器解決方案，它適合要求高達近50 kHz激勵頻率的應用。 </p><p&g

51、t;　　傳感器工作原理：通過傳感器的激勵頻率信號會根據(jù)傳感器的L或C瞬時值表現(xiàn)出相應的幅度、頻率或者相位的改變。例如，超聲波液流計會表現(xiàn)出相位偏移，而接近傳感器會引起幅度改變。 跟蹤這種變化阻抗的最常用方法就是監(jiān)視電路的諧振頻率。諧振頻率就是電容值等于電感值所在的頻率點。這也是頻率曲線上最大阻抗值對應的頻率點。在正常情況下，例如在靜態(tài)條件下，傳感器的L，R和C都具有一個唯一值，在諧振頻率Fo處具有最大阻抗值。當一個運動物體接近傳感器時，

52、那么傳感器的L和C值就會改變，并且產(chǎn)生一個新的諧振頻率。通過監(jiān)測諧振頻率的變化（從而導致阻抗的變化），就有可能推測出運動物體相對傳感器的移動距離。 </p><p>　　計算諧振頻率：計算電路的諧振頻率需要測量頻率和阻抗的關(guān)系，尤其是需要一個能夠在一定頻率范圍內(nèi)具有掃描能力的波形發(fā)生器。一種簡單、低成本的實現(xiàn)方法就是采用AD5930波形發(fā)生器。AD5930具有在一組預設(shè)置的頻率范圍內(nèi)提供線性掃描的能力。一旦條件設(shè)

53、定，就無需進一步的控制，除了一個用于啟動頻率掃描的觸發(fā)器。 </p><p>　　AD5930具有許多優(yōu)點：輸出頻率的分辨率為28 bit，所以用戶能以小于0.1 Hz的控制精度輸出頻率。其輸出頻率范圍為0～10 MHz，從而對選擇傳感器具有很大的靈活性。例如，有些傳感器的頻率范圍很窄，但是要求在此頻率范圍內(nèi)具有很高的分辨率。還有些傳感器可能需要很寬的調(diào)頻范圍，但是分辨率要求較低。采用這種方法很容易計算出傳感器的

54、諧振頻率。 </p><p>　　系統(tǒng)框圖：通過BF-535 DSP處理器設(shè)置AD5930數(shù)字波形發(fā)生器。需要對從AD5930產(chǎn)生的正弦波輸出電壓波形進行低通濾波和放大以便消除主時鐘（MCLK）、鏡像頻率和高頻噪聲產(chǎn)生的饋通。經(jīng)過濾波的信號可用作傳感器的激勵頻率源。根據(jù)傳感器的阻抗響應信號可能需要進行放大以便使其進入模數(shù)轉(zhuǎn)換器（ADC）的動態(tài)范圍內(nèi)。傳感器的輸出和激勵頻率源都輸入到AD7266一種12 bit、2

55、 MSPS的同步采樣雙ADC。將ADC輸出的數(shù)據(jù)保存在存儲器中以便做進一步的分析以計算出傳感器的相位和幅度偏移。完整的集成傳感器解決方案。 </p><p>　　上面介紹的分立解決方案是一種常用的傳感器阻抗測量解決方案。該方案可能需要許多分立元件，所以是一種高成本的傳感器分析解決方案。這些單獨的元件還會增加自身的誤差源。設(shè)計中的有源元件還會增加相位誤差，這也是需要校正。另外，還需要DSP處理一些復雜的數(shù)學計算，這

56、樣可能需要外部存儲器來存儲原始的ADC數(shù)據(jù)，從而會進一步增加成本。 </p><p>　　解決上述低頻率傳感器分析問題的解決方案是AD5933/4器件，它將上述主要處理模塊都集成到一顆芯片中。該芯片的內(nèi)核包括3個主要單元：用于提供頻率掃描的直接數(shù)字頻率合成器（DDS）波形發(fā)生器； 用于測量傳感器的響應的12 bit、1 MSPS ADC；以及最后能夠?qū)DC測量數(shù)據(jù)進行1024點離散傅立葉變換（DFT）運算的DS

57、P引擎。 </p><p>　　DFT運算結(jié)果提供一個實部（R）和一個虛部（I）數(shù)據(jù)，從而可以方便地計算出阻抗。采用下面的公式很容易計算出阻抗的幅度和相位：為了確定實際的實數(shù)阻抗值Z(ω)，通常需要進行頻率掃描?？梢杂嬎愠雒總€頻率點的阻抗，從而可以得出一條頻率與幅度的關(guān)系曲線。這樣就很容易測量出100 Ω～20 MΩ范圍內(nèi)的阻抗。該系統(tǒng)允許用戶設(shè)置一個2 V峰峰值（PK-PK）的正弦信號作為外部負載的激勵頻率源。

58、輸出范圍還可以設(shè)置為1V，500 mV和200 mV。頻率分辨率可以達到27 bit（< 0.1 Hz）。 </p><p>　　實現(xiàn)頻率掃描：為了實現(xiàn)頻率掃描，用戶必須首先設(shè)置頻率掃描所需要的條件：需要一個起始頻率、頻率間隔和掃頻點數(shù)。然后需要一個啟動命令開始掃描。在每個掃描頻點，ADC先完成1024個采樣，然后進行DFT計算以便提供該波形的實部和虛部數(shù)據(jù)。此實部和虛部數(shù)據(jù)通過I2C接口以兩個16 bit

59、字形式提供給用戶。片內(nèi)DSP處理單元的優(yōu)點是用戶不必進行復雜的數(shù)學計算，也無需存儲ADC原始數(shù)據(jù)，只需提供兩個16 bit的數(shù)據(jù)。因此，它還允許選擇更便宜的DSP解決方案，因為大大降低了對最終處理能力的要求。</p><p>　　由于未經(jīng)標定的系統(tǒng)只能使用典型的靈敏度和偏移值將輸出電壓轉(zhuǎn)換為壓力，測得的壓力將產(chǎn)生如圖1所示的誤差。 </p><p>　　這種未經(jīng)標定的初始誤差由以下幾個部分

60、組成： </p><p>　　1.偏移量誤差：由于在整個壓力范圍內(nèi)垂直偏移保持恒定，因此變換器擴散和激光調(diào)節(jié)修正的變化將產(chǎn)生偏移量誤差。</p><p>　　2.靈敏度誤差：產(chǎn)生誤差大小與壓力成正比。如果設(shè)備的靈敏度高于典型值，靈敏度誤差將是壓力的遞增函數(shù)。如果靈敏度低于典型值，那么靈敏度誤差將是壓力的遞減函數(shù)。該誤差的產(chǎn)生原因在于擴散過程的變化。 </p><p>

61、;　　3.線性誤差：這是一個對初始誤差影響較小的因素，該誤差的產(chǎn)生原因在于硅片的物理非線性，但對于帶放大器的傳感器，還應包括放大器的非線性。線性誤差曲線可以是凹形曲線，也可以是凸形曲線。 </p><p>　　4.滯后誤差：在大多數(shù)情形中，滯后誤差完全可以忽略不計，因為硅片具有很高的機械剛度。一般只需在壓力變化很大的情形中考慮滯后誤差。</p><p>　　標定可消除或極大地減小這些誤差，

62、而補償技術(shù)通常要求確定系統(tǒng)實際傳遞函數(shù)的參數(shù)，而不是簡單的使用典型值。電位計、可調(diào)電阻以及其他硬件均可在補償過程中采用，而軟件則能更靈活地實現(xiàn)這種誤差補償工作。 </p><p>　　一點標定法可通過消除傳遞函數(shù)零點處的漂移來補償偏移量誤差，這類標定方法稱為自動歸零。 </p><p>　　偏移量標定通常在零壓力下進行，特別是在差動傳感器中，因為在標稱條件下差動壓力通常為0。對于純傳感器，

63、偏移量標定則要困難一些，因為它要么需要一個壓力讀取系統(tǒng)，用以測量其在環(huán)境大氣壓力條件下的標定壓力值，要么需要獲取期望壓力的壓力控制器。 </p><p>　　差動傳感器的零壓力標定非常精確，因為標定壓力嚴格為0。另一方面，壓力不為0時的標定精確度取決于壓力控制器或測量系統(tǒng)的性能。</p><p><b>　　選擇標定壓力 </b></p>&l

64、t;p>　　標定壓力的選取非常重要，因其決定了獲取最佳精度的壓力范圍。實際上，經(jīng)過標定后實際的偏移量誤差在標定點處最小并一直保持較小的值。因此，標定點必須根據(jù)目標壓力范圍加以選擇，而壓力范圍可以不與工作范圍相一致。</p><p>　　為了將輸出電壓轉(zhuǎn)換為壓力值，由于實際的靈敏度往往是未知，因此在數(shù)學模型中通常采用典型靈敏度進行單點標定。</p><p>　　紅色曲線表示進行偏移量標

65、定（PCAL=0）后的誤差曲線，可以發(fā)現(xiàn)誤差曲線相對于表示標定前誤差的黑色曲線產(chǎn)生了垂直偏移。 </p><p>　　這種標定方法與一點標定法相比要求更為嚴格，實現(xiàn)成本也更高。然而與一點標定法相比，該方法可顯著提高系統(tǒng)的精度，因為該方法不僅標定了偏移量，還標定了傳感器的靈敏度。因此在誤差計算中可以使用靈敏度實際值，而非典型值。</p><p>　　綠色曲線表示精度提高。在這里，標

66、定是在0至500兆巴（滿標度）條件下進行。由于在標定點上誤差接近于0，因此為了在期望的壓力范圍內(nèi)得到最小的測量誤差，正確地設(shè)定這些點就顯得尤為重要。 </p><p>　　某些應用中要求在整個壓力范圍內(nèi)保持較高的精確度。在這些應用中，可以采用多點標定法來得到最理想的結(jié)果。在多點標定法中，不僅考慮了偏移量和靈敏度誤差，還考慮了大部分的線性誤差，紫紅色曲線所示。這兒用的數(shù)學模型與每個標定間距（在兩個標定點之

67、間）的兩級標定完全一樣。    </p><p>　　如前所述，線性誤差具有一致的形式，且誤差曲線符合二次方程的曲線，具有可預測的大小和形狀。對于未采用放大器的傳感器更是如此，因為傳感器的非線性從本質(zhì)上是基于機械原因（這是由硅片的薄膜壓力引起）。 </p><p>　　線性誤差特性的描述可以通過計算典型實例的平均線性誤差，確定多項式函數(shù)（a

68、×2+bx+c）的參數(shù)而得到。確定了a、b和c后得到的模型對于相同類型的傳感器都是有效的。該方法能在無需第3個標定點的情況下有效地補償線性誤差。 </p><p>　　摩托羅拉MPX2300的補償實例，MPX2300是一種主要應用于血壓測量的溫度補償傳感器。多項式模型可由10個傳感器的平均線性誤差得到，補償后的誤差約為最大初始線性誤差的十至二十分之一，如圖3虛線所示。該誤差補償方法只需兩點標定